The Design And Implementation Of Different Enterprise Fundamental Forecasting Systems Based On Financial Data

In recent years,Internet + finance has become the development trend of financial forecasting.However,some companies exaggerate the corporate earnings forecasts of their self-disclosure,making it difficult for investors to understand the true financial situation of enterprises.Therefore,it is crucial to help investors predict the data of the next quarter from the financial data of the past years and avoid being confused by the company.The focus of fundamental forecasting is earnings forecasts,which can directly reflect the company's current operations.Research shows that China's profit forecast is generally carried out according to specific industries,which leads to a wide attention from all walks of life to the industry with higher incomes,while industries with small business values didn't get enough attention,resulting in the lack of industry forecasting models.The existing research is mainly based on a single industry,and the differences in financial data characteristics among different industries and the financial data models applicable in different industries remain to be studied.This article has conducted the following research on these issues:First of all,this paper builds the financial data of 61 industries based on the financial statements of 3,000 different listed companies,and then studies whether there are differences in the characteristics of financial data among different industries.This paper collects and analyzes the financial data after construction,and then judges whether there are characteristics unique to each industry,and finds the correlation between these characteristics and profit.Finally,the feature with higher relevance is used as the feature of subsequent model selection.Secondly,this thesis studies seven kinds of machine learning methods,such as long-term and short-term memory networks,and a combination of three normalization methods(Hybrid optimization,H-op for short),which explores whether different industries should adopt different normalization methods and different machine learning algorithms after constructing industry characteristics.In addition,this papaer will present the best prediction methods for each industry.H-op trains different industry data,compares forecasting methods according to financial industry forecasting evaluation criteria,and selects the optimal model as the current industry forecasting model.Experiments show that the proposed machine learning and multiple normalized combinations can get the current optimal solution,and the prediction accuracy can reach more than 96%.Finally,this paper uses the above research conclusions to set up a fundamental prediction system of different enterprises based on financial data.This paper will use H-op to select the best model of the current industry and the corresponding normalization method to predict the financial situation of different companies in the industry in the next quarter.Then we will combine other financial data to make judgments and assessments on the risks and vitality of the company,and directely dispaly these information.In this paper,the specific design and implementation of each module are given,and the function and performance of the system are tested.The results show that the system implemented in this paper has good prediction and display functions.